An IC yield enhancement approach by ARMA modeling and dynamic process control

Abstract

The IC industry has been growing rapidly in the past decades. The continuous scaling-down of the feature size requires IC machines of highest performances, and pushes the IC manufacture to its utmost technology limits. Nowadays, IC manufacturers employ tightly fixed process parameters as their strategy to improve the yield. In this paper, a "softer" way is proved to be more potential in further improving and managing the yield of IC products. A novel concept which suggests running an IC procedure with dynamic process parameters is explained in detail. By using auto-regressive moving-average (ARMA) model, effects of the key process steps can be predicted and the related process parameters can be adjusted self-reactively. Since the discussion indicates distinct improvements of the IC yields, it seems the time to break the rule of "using stationary process parameters".

Adaptive Dosing Control System Through ARIMA Model for Peristaltic Pumps

“…Once p reaches 10 or higher, the auto-correlation patterns become similar. Since further increases in p require higher computation time without significant improvement in terms of auto-correlation, AR (10) was chosen as the optimal parameter for further analysis.…”

“…Based on what was discussed in Section III-B1, an AR (10) model was chosen for these experiments. The workstation used was exactly the same as described in Section II-A, the overall control system is depicted in Figure 16, and the logic representation of the ADCS is shown in Figure 13.…”

“…Otherwise, the next step is performed. 3) Starting from the TW previously observed volumes in V dispensed , an AR (10) model is used to forecast the next N fillings, which are stored in the F volumes [N ] array, as shown in Figure 17b. 4) For each predicted quantity V p in F volumes [N ]: a) To increase the accuracy of the PP, volume compensation is performed according to the following formula:…”

“…1) The AR (10) model is used to forecast the next filling volume, denoted as V p , based on the previous TW observations (Figure 19b).…”

“…In addition to milling machines, other manufacturing processes have been proposed, such as integrated circuit production [10], precision multi-axis motion control systems with contour performance orientation [11], and additive manufacturing for thermal distortion compensation [12]. Furthermore, recent research has explored the application of different control techniques in various domains, including complex systems [13] and adaptive control [14], [15].…”